ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.204
Committed: Tue Jan 23 15:55:45 2007 UTC (17 years, 3 months ago) by root
Branch: MAIN
Changes since 1.203: +3 -0 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 Revision history for Perl extension Coro.
2
3 TODO: better (earlier) error messages on broken transfer's
4 TODO: destroy gets only called on cancel: fix(?)
5 TODO: fast reloads in cf cause segfaults (coro_state_free->destroy->crash)
6
7 - do AnyEvent model detection earlier, avoiding problems
8 caused by first using AnyEvent and later Coro::Event.
9
10 3.41 Mon Jan 22 19:19:49 CET 2007
11 - readline on Coro::Handle did not support undefined $/,
12 nor did it deliver partial lines on EOF or error.
13 - implement malloc fallback for stack allocation because
14 stupid broken idiotic OSX has a stupid broken
15 idiotic fits-the-whole-os mmap "implementation" and
16 my dick feels longer if Coro is portable even to
17 obsolete platforms.
18
19 3.4 Fri Jan 19 21:52:54 CET 2007
20 - remove t/09_timer.t, as it isn't really testing much
21 but was rather flaky in practise.
22 - async_pool coro would keep arguments and callback alive until
23 it was reused.
24 - cancellation of a coroutine could cause spurious idle calls
25 in cede_notself.
26
27 3.3 Sat Jan 6 03:45:00 CET 2007
28 - implement $coro->on_destroy.
29 - Coro::Event blocking semantics have been changed,
30 documented and - hopefully - improved.
31 - fix nice adding, not subtracting, from priority.
32 - fix ->prio and api_is_ready (patch by Mark Hinds).
33 - fixed an assert ("... == prev__cctx->idle_te")
34 that could errronously trigger.
35 - fix various large and small memleaks.
36 - use a (hopefully) more stable cancel implementation
37 that immediately frees the coroutine data.
38 - cede/cede_notself return a status now.
39 - added Coro::guard function.
40 - added a global coroutine pool for jobs (on my machine,
41 I can create and execute 48k simple coros/s with async,
42 and 128k coros with async_pool).
43 - Coro::AIO now uses the coroutine priority as io priority.
44
45 3.2 Fri Dec 22 05:07:09 CET 2006
46 - improve portability to slightly older perls.
47 - use cleaner coroutine destruction.
48 - simplify configuration for users.
49 - optionally (unrecommended) prefer perl functions over
50 their coro replacements.
51
52 3.11 Tue Dec 5 13:11:24 CET 2006
53 - fixed some bogus assert's, but as perl.h disables assert even
54 without NDEBUG (thank you very much), not too many people should
55 notice (that did include myself). Andreas König noticed, though :)
56 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
57 it. Also noticed by Andreas König :)
58 - save/restore tainted status.
59 - verified to pass the testsuite on my 5.9.5.
60
61 3.1 Mon Dec 4 23:03:40 CET 2006
62 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
63 - incompatible change: transfer flags are now per-state.
64 - give a better error message on deadlock.
65 - document Coro::nready.
66 - enhanced testsuite.
67
68 3.01 Sun Dec 3 23:47:42 CET 2006
69 - forgot to include Coro::Timer.
70
71 3.0 Sun Dec 3 22:57:25 CET 2006
72 - the "FINALLY COMPLETELY STABLE" release (coming soon:
73 the "FAMOUS LAST WORDS" release).
74 - implement a new stack sharing algorithm, which uses a stack
75 pool (size currently hardcoded).
76 - make stack sharing mandatory (it no longer uses a heuristic).
77 - eval/die no longer cause weird problems under heavy use.
78 - Coro::Event could cause livelocks if it was used but
79 no Coro::Event watchers were used.
80 - Coro::Event now uses asynccheck as prepare does not
81 check for changed watchers.
82 - Coro::Event allows multiple waiting coros and will wake up one
83 per event.
84 - Coro::Event should be cleaner and more efficient now.
85 - new utility function Coro::unblock_sub.
86 - document the sad fact that Event is no longer reentrant.
87 - putting a coroutine into the ready queue twice could under
88 some circumstances lead to stack corruption.
89 - minor incompatibility: subclassing Coro::State is supported
90 directly now without going through a _coro_state member.
91 - state/coro switching is much faster now.
92 - very minor optimisations and code/documentation cleanup.
93 - avoid problems due to compiler inlining.
94 - removed timers from Coro::Timer -> use AnyEvent instead.
95 - replaced Coro::idle coroutine by (cleaner) idle handler.
96 - updated to newest libcoro.
97 - implement enhanced support for valgrind.
98 - implement is_ready and return value for ready.
99 - removed Coro::Cont, it was a misnomer (it's generators),
100 and it was somewhat annoying to get it right. Will come back
101 if somebody asks for it :)
102 - many ->wait methods and Coro::Event could return
103 spuriously without the event having happened.
104
105 2.5 Tue Nov 7 12:22:33 CET 2006
106 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
107 use AnyEvent, moved them to Coro/.
108 - added Coro::LWP which contains all the uglyness required to
109 make LWP non-blocking.
110 - should work with perl 5.9.x now (Andreas König made me do it).
111 - fixed another bug in Coro::Select when the timeout was undef.
112 - reuse PL_start_env for all coros, saves some memory per coroutine.
113 - manage PL_top_env differently, hopefully to avoid panic: top_env.
114 - timeout argument was not properly used in Coro::Socket.
115 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
116 - emulate undocumented(!) functionality of IO::Socket required
117 by LWP(!!).
118 - updated eg/lwp to work with newer lwp's.
119 - remove "FATAL: uncaught exception" prefix. Coroutines that die
120 kill the whole process, just as exceptions in the main "coroutine"
121 did already.
122
123 2.1 Wed Nov 1 23:01:13 CET 2006
124 - fix a long-standing bug in Coro::Select where select with
125 zero timeout would instead change the current default filehandle.
126 - use a simpler and hopefully more robust way to clone padlists
127 (uses less memory and a perl function instead of our own).
128 - coro can now create a stack guard area on many architectures.
129 - Coro::AIO properly reexports additional functions from IO::AIO.
130 - updated libcoro with a workaround for OS X,
131 pach and testing by Michael Schwern.
132
133 2.0 Tue Oct 24 05:47:17 CEST 2006
134 - support additional aio requests in Coro::AIO.
135
136 1.9 (never properly released due to a glitch)
137
138 1.8 Thu Feb 2 00:59:06 CET 2006
139 - applied suggested patch by SAMV to avoid problems during stupid
140 mark & sweep gc run.
141 - applied patch by Scott Walters for 5.9.3 compatibility.
142
143 1.7 Tue Dec 27 01:41:58 CET 2005
144 - added Coro::AIO, a thin wrapper around IO::AIO.
145 - improved Makefile.PL explanations.
146
147 1.6 Mon Dec 12 21:30:05 CET 2005
148 - additionally save PM_curpm might fix as-of-yet
149 unseen problems with regex matches being attributed to the
150 wrong package.
151 - add t/10_bugs.t, which currently checks against imho broken
152 perls that use 0.26MB of stack space per Perl_magic_get
153 invocation (newer linuxes) as opposed to the 0.0002MB perl
154 normally uses.
155 - make stacksize configurable for the ultimate debian experience.
156
157 1.51 Mon Dec 12 18:48:36 CET 2005
158 - remove debugging warn() accidentally left in Coro::Select.
159
160 1.5 Tue Nov 29 13:32:44 CET 2005
161 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
162 - fix Coro::Select 'select'.
163 - strict'ify some modules.
164 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
165 - use XSLoader in selected modules.
166 - remove some 5.6 compatibility cruft.
167
168 1.4 Tue Sep 6 00:11:05 CEST 2005
169 - libcoro did not take into account the trampoline on amd64,
170 when 'l'inux method was used.
171
172 1.31 Tue Aug 30 23:31:33 CEST 2005
173 - some portability fixes/workarounds.
174
175 1.3 Sat Aug 20 03:08:56 CEST 2005
176 - no code changes. module seems to work fine.
177 - improve Coro::State docs, remove reference to nonexisting
178 Coro::State::flush method.
179 - no longer autodetect windows, present it as an option instead.
180
181 1.2 Mon May 16 02:00:55 CEST 2005
182 - included libcoro.c earlier in State.xs, might improve portability.
183 - use a faster and possibly more stable Coro::Cont implementation.
184 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
185 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
186 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
187
188 1.11 Thu Mar 3 18:00:52 CET 2005
189 - change of contact address.
190
191 1.1 Tue Feb 22 20:51:16 CET 2005
192 - support [l]inux method on amd64.
193 - allow some context switches while compiling/eval'ing.
194
195 1.0 Tue Aug 10 03:47:27 CEST 2004
196 - Event 0.89 fixes the bug again.
197 - fix the remaining(??) memleak. I didn't free the cache-AV for
198 padlists, so one AV per created anon-sub leaked.
199
200 0.97 Fri May 14 15:23:32 CEST 2004
201 - remove stupid gcc extension that was an experiment only anyway.
202
203 0.96 Thu May 13 18:09:29 CEST 2004
204 - tell people if they have yet another broken Event variant.
205 - made Coro to work _better_ with threads (it seems that, as long
206 as you use Coros in one thread only everything will work fine.
207 Looking at the perl source the problem might be missing locking
208 between Coro and some perl internal routines. That's not easily
209 fixable).
210 - (hopefully) got rid of the static xnext inside transfer.
211
212 0.95 Sun Feb 15 03:20:28 CET 2004
213 - removed Coro::State::flush, implemented a new and much
214 faster caching mechanism by attaching magic to the CV.
215 - a memleak with (real) closures remains.
216 - dramatically reduced size of stacks to 65536 (linux/x86), after
217 observing that even myhttpd never had more than a kilobyte
218 of stack in use (stack is still only physically allocated in
219 pagesize increments for systems suporting mmap).
220 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
221 - fix for threaded perl.
222
223 0.9 Sun Nov 23 04:32:39 CET 2003
224 - fix a few very hard to track down but (of course) stupid
225 memory leaks.
226 - stringify version number to avoid locale problems :(.
227 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
228 hopefully ignoring it "just works".
229
230 0.8 Wed Nov 5 19:38:40 CET 2003
231 - port to perl5.005_03, but only 5.8.x+ is supported!
232 - honor LocalAddr even without LocalPort in Coro::Socket.
233
234 0.7 Tue May 27 03:12:38 CEST 2003
235 - the version jump indicates some level of testing, not gobs
236 of new features.
237 - uh... I found the prompt function in ExtUtils::MakeMaker.
238 Highly correct stuff, that is...
239 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
240 the fix is not well understood by the author, so beware :(.
241
242 0.652 Thu May 8 02:54:46 CEST 2003
243 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
244 cause version 4 doesn't have ucontext.
245 - Benjamin Reed reported that setjmp works fine on darwin,
246 so preselect it.
247
248 0.651 Sat Mar 29 15:00:23 CET 2003
249 - fix a bug in Coro::Handle where some bytes could get lost
250 on reading, reported by jason@nichego.net.
251
252 0.65 Sun Mar 23 00:08:26 CET 2003
253 - added a README.linux-glibc.
254 - new module Coro::Select.
255 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
256 (I never claimed that I know what I am doing ;).
257 - set default to setjmp/longjmp on non-x86-linux.
258
259 0.6 Thu Nov 21 11:09:06 CET 2002
260 - made lazy context switching the default.
261 - fixed the bug where SemaphoreSet::guard
262 was the same as timed_guard and timed_guard was missing.
263 - fix a memleak in Coro::Timer.
264 - reclassified context sharing as not experimental.
265
266 0.534 Sun Apr 14 03:05:12 CEST 2002
267 - fixed backspace => tab in header parsing (myhttpd).
268 - added eg/readline, for elmex.
269 - Coro::Event now calls ->start on first call instead of
270 ->again, which created wrong timeouts for the first ->next.
271 - fixed a bug where Coro::Socket returned a closed socket instead
272 of nothing on ECONNREFUSED and similar error conditions.
273
274 0.533 Mon Feb 18 18:49:40 CET 2002
275 - adapt to finally working Event-0.85.
276 - fixed buggy event initialization (trapped under windows, but how
277 could it possibly have worked anywhere?)
278
279 0.532 Wed Jan 16 02:45:32 CET 2002
280 - removed the reference to pp_entersub, might work on
281 (some) win32 perls now (testsuite works under cygwin,
282 without Event, which freezes).
283
284 0.531 Mon Dec 10 22:18:44 CET 2001
285 - Fixed a bug in SemaphoreSet::timed_down.
286
287 0.53 Tue Nov 27 21:11:13 CET 2001
288 - some tests for Event.
289 - slightly different internal architecture to get refcnt into the object
290 and not the reference. might fix a few bugs, certainly did introduce
291 new ones.
292 - Coro::Timer (independent of Event).
293 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
294
295 0.52 Tue Nov 6 21:36:18 CET 2001
296 - ported to cygwin (trivial).
297 patch by Gerrit P. Haase <gp@familiehaase.de>.
298 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
299
300 0.51 Thu Nov 1 20:39:01 CET 2001
301 - terminate/cancel now work properly (otherwise termination
302 could cause a "next coroutine is not and contains not..."
303 error.
304 - added Coro::Socket::shutdown.
305 - Coro::Event::loop no is the same as Event::loop.
306 - implemented terminate with args + join.
307
308 0.5 Fri Sep 28 16:15:35 CEST 2001
309 - fixed "print" on a Coro::Handle. The print method worked.
310 - small tweaks (seem to reduce memory consumption a lot)
311 in various modules.
312 - splendid use of "no warnings" scattered throughout.
313 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
314
315 0.49 Sun Sep 16 02:42:45 CEST 2001
316 - changed some method calls to function calls for speed
317 inside Coro::Handle.
318 - make Coro::Handle use an array instead of a hash for
319 speed reasons.
320 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
321 sjlj and ucontext should now work.
322 - IRIX-specific port for libcoro.
323 - swapped order of accept results to match IO::Socket.
324 - changed getsock/peername to sock/peername to match IO::Socket.
325 - fixed a bug that caused segfault when returning to main task
326 under some circumstances.
327 - other bugfixes.
328
329 0.45 Sun Sep 2 02:54:01 CEST 2001
330 - new method Coro::Handle::timeout.
331 - corrected speling of set...name to get...name in Coro::Socket.
332 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
333 - Coro::Handle now supports fileno and FILENO.
334 - added eg/myhttpd, a web server.
335 - Coro::Socket now accepts numeric ports as well as "port(number)"
336 syntax.
337 - moved some scheduling primitives into xs code. more to come.
338 - new simple priority system.
339 - implemented Coro::Event using XS, almost four times faster!
340 - small memory corruption problem fixed (boy that was difficult).
341
342 0.13 Wed Aug 8 16:53:07 CEST 2001
343 - new method Coro::Handle::readline.
344 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
345 (see eg/lwp on how to make LWP non-blocking).
346 - renamed Coro::Socket::new_inet to new.
347 - added Coro::Util, some utility functions.
348 - die/eval now works better (still get annoying (but true ;)
349 "Callback called exit" messages).
350 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
351 - much better Coro::Socket support.
352
353 0.12 Fri Jul 27 04:19:01 CEST 2001
354 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
355 - very experimental stack sharing algorithm. not 100% safe but
356 should work well in practise ;)
357 - added Coro::RWLock.
358 - Coro::Specific now works.
359
360 0.11 Tue Jul 24 22:49:21 CEST 2001
361 - added specialized hack for newer and older linux versions (fast).
362 - renamed Coro::Event::IO to Coro::Handle.
363 - new module Coro::Socket.
364
365 0.1 Tue Jul 24 01:47:53 CEST 2001
366 - release candidate 3. A rather new internal structure :(
367 - the great renaming: Damian Conway gave me a suitable replacement
368 for yield.
369 - added Coro/libcoro, a portable coroutine implementation for C
370 (not even perl-dependent!!!), which can be used by Coro::State.
371 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
372 perl-5.6.
373 - new function Coro::Event::idle.
374 - the idle process is now overriden
375 by default in Coro::Event.
376 - Coro::Channel now enforces the size.
377 - canceling events now works in all cases.
378 - Coro state now include $_ and $@.
379 - yet another bug workaround that I do not really understand :(
380 - new module Coro::Event::IO (very undocumented) to do
381 non-blocking i/o.
382 - performance tuning.
383
384 0.08 Thu Jul 19 06:13:25 CEST 2001
385 - release candidate 2 (new functionality).
386 - add Coro::State::flush function.
387 - transfer now supports a flags value (mostly for speed ATM).
388 - might compile and work in the presence of threads now.
389 - continuations have a saner syntax.
390 - no more memleaks.
391
392 0.07 Tue Jul 17 17:40:18 CEST 2001
393 - release candidate 1 ;)
394 - slightly nicer code.
395 - fixed a scheduling bug in Coro::Event.
396
397 0.06 Tue Jul 17 04:23:24 CEST 2001
398 - ok, I found the showstopper - the same sub must not be
399 re-entered in two different coroutines, otherwise => crash. I
400 see no easy solution to this problem, except by walking the call
401 chaing and saving/restoring the cv's, which is what I do now.
402 - memory leaks still latent, especially at thread termination.
403 - Coro::Event now works (haha).
404
405 0.05 Sun Jul 15 17:32:20 CEST 2001
406 - fixed a few issues in Coro::Event.
407 - I forgot to include Coro::Event in 0.04 :(:(:(
408
409 0.04 Sun Jul 15 05:24:59 CEST 2001
410 - @_ is now properly localized.
411 - Coro::State is now easier subclassable.
412 - Coro::Cont now coroutine-aware.
413 - Coro::Specific is a low-overhead module to create
414 coroutine-specific vars.
415 - Coro::Event provides a simple interface to Event.
416
417 0.03 Fri Jul 13 14:51:52 CEST 2001
418 - transfer() now implemented in XS (beware).
419 - new module Coro::Cont for really faked continuations.
420 - big internal architecture changes: Coro::State is now
421 really low-level and can thus be used to implement other
422 interesting things, While "Coro::" implements a process-like
423 model. Still crude and subject to change.
424 - $_ and $@ are no longer being localized.
425
426 0.02 Tue Jul 10 01:38:17 CEST 2001
427 - implemented "async" attribute.
428 - $_ and $@ are now localized.
429 - added Coro::Channel.
430 - more testcases, still no docs.
431
432 0.01 Tue Jul 3 02:18:41 CEST 2001
433 - original version; copied from Convert::Scalar.
434